Thermodynamic and Aerosol Controls in Southeast Pacific Stratocumulus 1250 D B. M

advertisement
1250
JOURNAL OF THE ATMOSPHERIC SCIENCES
VOLUME 69
Thermodynamic and Aerosol Controls in Southeast Pacific Stratocumulus
DAVID B. MECHEM
Atmospheric Science Program, Department of Geography, University of Kansas, Lawrence, Kansas
SANDRA E. YUTER
Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, North Carolina
SIMON P. DE SZOEKE
College of Oceanic and Atmospheric Sciences, Oregon State University, Corvallis, Oregon
(Manuscript received 20 June 2011, in final form 20 November 2011)
ABSTRACT
A near-large-eddy simulation approach with size-revolving (bin) microphysics is employed to evaluate the
relative sensitivity of southeast Pacific marine boundary layer cloud properties to thermodynamic and aerosol
parameters. Simulations are based on a heavily drizzling cloud system observed by the NOAA ship Ronald H.
Brown during the Variability of the American Monsoon Systems (VAMOS) Ocean–Cloud–Atmosphere–Land
Study—Regional Experiment (VOCALS-Rex) field campaign. A suite of numerical experiments examines the
sensitivity of drizzle to variations in boundary layer depth and cloud condensation nuclei (CCN) concentration
in a manner consistent with the variability of those parameters observed during VOCALS-Rex. All four simulations produce cellular structures and turbulence characteristics of a circulation driven predominantly in
a bottom-up fashion. The cloud and subcloud layers are coupled by strong convective updrafts that provide
moisture to the cloud layer. Distributions of reflectivity calculated from model droplet spectra agree well with
reflectivity distributions from the 5-cm-wavelength scanning radar aboard the ship, and the statistical behavior
of cells over the course of the simulation is similar to that documented in previous studies of southeast Pacific
stratocumulus. The simulations suggest that increased aerosol concentration delays the onset of drizzle, whereas
changes in the boundary layer height are more important in modulating drizzle intensity.
1. Introduction
Recent studies have focused on the representation of
low clouds as a critical component leading to uncertainties
in climate change projections simulated by large-scale
models (Bony and Dufresne 2005; Medeiros et al. 2008).
The climatological importance of large areas of marine
boundary layer clouds over the eastern subtropical oceans
is well known (Klein and Hartmann 1993). The sensitivity
of the global radiation budget to joint interactions of low
clouds and aerosol has traditionally been cast into the
conceptual framework of the albedo (Twomey 1974,
1977) and cloud lifetime (Albrecht 1989) indirect effects.
Corresponding author address: David B. Mechem, Atmospheric
Science Program, Department of Geography, University of Kansas, 1475 Jayhawk Blvd., 213 Lindley Hall, Lawrence, KS 660457613.
E-mail: dmechem@ku.edu
DOI: 10.1175/JAS-D-11-0165.1
Ó 2012 American Meteorological Society
Some progress has been made in observationally quantifying the Twomey effect (e.g., Feingold et al. 2003, 2006),
although care is required in constraining the analysis for
constant liquid water path (LWP). Quantifying the second indirect effect has proved more problematic because
of the complicated and poorly understood feedbacks
among aerosol, cloud, and precipitation processes. This
study focuses on a subset of low clouds, marine stratocumulus clouds, which are climatologically important
(Klein and Hartmann 1993; Bony and Dufresne 2005)
and also difficult for current global climate models to
represent correctly (Wyant et al. 2010).
Although often thought of as horizontally homogeneous, marine stratocumulus cloud properties are, in
reality, highly variable in space and time. Particularly
noticeable are pockets of open cells (POCs; Stevens
et al. 2005b), which are regions characterized by anomalously low aerosol concentration, low cloud fraction, and cells of small area with high precipitation rates
APRIL 2012
MECHEM ET AL.
($20 mm day21). Collection of cloud droplets associated with the precipitation process reduces the total
droplet number, and ultimately cloud condensation nuclei (CCN) (Hudson 1993; Garrett and Hobbs 1995;
Feingold et al. 1996; Mechem et al. 2006; Wood 2006).
Work employing numerical models has suggested the
conceptual model of a positive feedback loop, whereby
drizzle scavenges CCN, leading to enhanced drizzle
production, which further scavenges CCN (Feingold and
Kreidenweis 2002). Conceptual models have generally
assumed a number of dynamical feedbacks that can ultimately lead to cloud thinning and cloud field breakup
(Albrecht 1989; Paluch and Lenschow 1991), with recent
modeling studies suggesting the presence of precipitation
as one mechanism leading to the formation of POCs
(Savic-Jovcic and Stevens 2008; Wang and Feingold
2009a,b). Precipitation, however, is not always accompanied by reduced cloud fraction (Stevens et al. 1998;
Ackerman et al. 2009).
The 2001 East Pacific Investigation of Climate (EPIC)
field campaign took place over the southeast Pacific (SEP)
region off the coast of South America (Bretherton et al.
2004). The research cruise sailed from the Galapagos to
Chile and found that thicker, nocturnal, drizzling clouds
were generally associated with more pronounced mesoscale variability relative to their daytime counterparts.
Specifically of interest in EPIC were the transition regions between solid stratocumulus and broken cloud
fields, as these were thought to denote the focal point for
mechanisms important in driving mesoscale variability.
Employing data from scanning C-band radar, Comstock
et al. (2005, 2007) found that the strongest drizzle rates
lay at the transitions between closed and open mesoscale
cellular POC structures. Individual drizzle cells drew
moisture from the surface layer and had lifetimes up to
2 h (Comstock et al. 2005). Mixed-layer budgets calculated from observations during EPIC showed that drizzle
has a negligible effect on the mean water budget since
most of the drizzle evaporates before reaching the ocean
surface (Caldwell et al. 2005). However, by stabilizing the
subcloud layer, drizzle promoted decoupling of the cloud
and subcloud layers by suppressing turbulent transports
below cloud base.
The Variability of the American Monsoon Systems
(VAMOS) Ocean–Cloud–Atmosphere–Land Study—
Regional Experiment (VOCALS-Rex, which we shall
refer to simply as VOCALS from here forward) field
campaign took place during October–November 2008.
VOCALS was largely driven by the recognition of global
climate model (GCM) uncertainties and biases over the
tropics and subtropics arising from inadequately representing the coupled atmosphere–ocean–land system. The
experiment involved an extensive suite of airborne and
1251
ship-based observational platforms in order to characterize the mechanisms that drive the variability in the
SEP climate system (Wood et al. 2011).
During VOCALS, the National Oceanic and Atmospheric Administration (NOAA) ship Ronald H. Brown
(RHB) was equipped with remote sensing instruments
capable of characterizing cloud, precipitation, and atmospheric properties. A scanning C-band (5-cm) radar
sampled the three-dimensional radar reflectivity and radial velocity structures of the drizzle field within a 60-km
radius from the ship, and a profiling 35-GHz Doppler
cloud radar observed the cloud field just above the ship.
The scanning 5-cm radar has been shown to be highly
useful for documenting the 3D spatial structure and
temporal evolution of the drizzle field (Yuter et al. 2000;
Comstock et al. 2005, 2007).
One unexpected and significant finding from the Cband radar aboard ship during VOCALS was the occasional occurrence of echo regions with radar reflectivity
greater than 30 dBZ. The reflectivity values were unusually high for boundary layer clouds, which tend to
have peak reflectivities in the 0–25-dBZ range [see reflectivities in Frisch et al. (1995)—Atlantic Stratocumulus
Transition Experiment (ASTEX), northeast Atlantic;
Stevens et al. (2003)—the Second Dynamics and Chemistry of Marine Stratocumulus field study (DYCOMS-II),
northeast Pacific; and Bretherton et al. (2004)—EPIC
and SEP]. Figure 1 illustrates an example that includes
one of these unusually strong precipitation cells. Infrared
satellite imagery indicates the radar sampled a region of
transition from solid to broken cloud (Figs. 1a,b). The
vertical cross section [range–height indicator (RHI)] in
Fig. 1c taken to the northeast of the radar indicates drizzle
cells with a maximum reflectivity of about 20 dBZ. A line
of large reflectivities was located to the southwest of the
ship, with the vertical cross section indicating reflectivity
values over 35 dBZ. The echo top reached nearly 2 km in
altitude, and its shape suggested an updraft penetrating
into the inversion zone.
What is the explanation for these large radar reflectivities? Numerical simulations based on one of the
VOCALS research flights found that drizzle and POC
formation were more sensitive to boundary layer moisture and temperature perturbations than to aerosol
number concentration (Wang et al. 2010). A comparison
of inversion height zi and water vapor mixing ratio
(averaged over the lowest 200 m) calculated from the
RHB soundings, stratified by broad drizzle rate categories, suggests that drizzle rates tend to be maximized
for boundary layer thermodynamic conditions that are
moister and deeper (Fig. 2). This perspective is consistent with observations from the NOAA Earth System
Research Laboratory (ESRL) profiling 35-GHz cloud
1252
JOURNAL OF THE ATMOSPHERIC SCIENCES
VOLUME 69
FIG. 2. Boundary layer depth zi as a function of mean (surface to
200 m) water vapor mixing ratio qy for the RHB soundings coinciding with drizzle occurrence in the radar sampling volume.
Soundings are stratified according to weak, intermediate, and
strong drizzle. Any detected meteorological echo (i.e., not sea
clutter) , 20 dBZ is termed weak drizzle. Intermediate drizzle
contains at least some pixels $ 20 dBZ. Periods of strong drizzle
contain at least 45 pixels $ 30 dBZ that last for at least 30 min.
tend to dominate the precipitation process, we acknowledge that the CCN measurements at the surface may not
be entirely representative of those at cloud base, particularly in cases where substantial evaporation of precipitation can lead to a strong vertical gradient in CCN
concentration. Figures 2 and 3 thus serve to motivate the
research; more observational evidence would be required
FIG. 1. Satellite and radar view of VOCALS cloud regime transition. (a) Infrared satellite imagery. The green circle represents radar coverage of the RHB. (b) Plan view of radar reflectivity from the
1.08 elevation scan of the C-band radar aboard the RHB. (c) Vertical
cross sections (RHIs) taken to the northeast and southwest of the
RHB, as indicated by the dashed line in (b).
radar aboard the RHB, which indicates that heavier precipitation is associated with deeper boundary layers and
thicker clouds [see Fig. 2 in de Szoeke et al. (2010b)].
Although the concentration of CCN observed by the
ship is weakly correlated with drizzle rate (Fig. 3) for
weak and intermediate cases, the strong drizzle cases
were accompanied both low and high concentrations of
CCN. All other conditions being equal, lower CCN
concentrations should favor stronger drizzle. Although
the observations suggest that factors apart from CCN
FIG. 3. RHB observations of CCN concentration (S 5 0.6%),
stratified according to drizzle regime as in Fig. 2. Each trace corresponds to 61.5 h from a sounding time. (CCN data courtesy
Dave Covert, University of Washington).
APRIL 2012
MECHEM ET AL.
to demonstrate concretely the relationship between drizzle and cloud-level CCN.
Both the thermodynamic analysis and the cloud radar
data suggest a robust relationship among boundary layer
depth (likely modulated by entrainment processes and
large-scale vertical velocity) and cloud thickness and
precipitation. The large-scale vertical velocity over the
southeast Pacific is predominantly subsidence, with a
superimposed ‘‘upsidence wave’’ (Garreaud and Muñoz
2004) arising from a gravity wave response to the diurnal
heating of the continent. The upsidence wave response is
complicated and depends on longitude [e.g., Fig. 5a in
Wood et al. (2009)], but it is largely in phase with the
diurnal cycle of cloud thickness over the SEP. Caldwell
and Bretherton (2009), however, found in multiday simulations of the EPIC cruise that the diurnal cycle of largescale vertical velocity (mostly subsidence) played little
role in modulating the LWP, suggesting that the diurnal
variability in radiative fluxes overwhelmed the effects of
the upsidence wave.
This paper employs a numerical simulation framework to explore the hypothesis that natural variability in
boundary layer thermodynamic properties exerts a
greater influence on cloud and precipitation processes
than does variability in aerosol concentrations. We do
this by varying the thermodynamic and aerosol properties in a way consistent with the environmental variability measured over the two RHB ship cruises during
the VOCALS campaign (de Szoeke et al. 2010b). Because of the complexities involved in longer-term feedbacks, our focus is on primary cloud and precipitation
responses that occur over the time frame from 8 to 12 h
of the simulation.
2. Methodology
Results from EPIC and preliminary analysis of the
VOCALS observations demonstrate that precipitation
and mesoscale organization go hand in hand. This suggests that the traditional small-domain large-eddy simulation (LES) approach is insufficient for representing
the mesoscale aspects of the cloud organization and
precipitation processes. Relatively few studies capture
with any fidelity both boundary layer turbulence and
mesoscale circulation, although the recent trend has
been toward employing larger and larger domains in
order to capture mesoscale aspects of the circulation.
Xue et al. (2008) employed a 12.4 3 12.4 km2 domain
and found a relationship between precipitation-driven
outflow and open cells accompanying trade wind cumulus. Savic-Jovcic and Stevens (2008) used typical LES
grid spacings (35 m in the horizontal) over a somewhat
larger domain of size 25.6 3 25.6 km2, whereas Wang
1253
and Feingold (2009a,b) coarsened the grid spacing to
300 m in order to run over a significantly larger (60 3
60 km2) domain. The trend in these studies has been to
recognize that the turbulent flow and the mesoscale
variability must both be accounted for in order to attain
a faithful representation of boundary layer cloud structures and evolution.
a. Near-LES numerical framework
In the spirit of these recent studies, we employ the
numerical framework of ‘‘near-LES,’’ the aim being to
conduct simulations with a large enough domain to
capture mesoscale structures but at sufficient resolution
to adequately represent the turbulent boundary layer
fluxes. This approach is also known as very-large-eddy
simulation (VLES), particularly in the engineering disciplines. Our simulations use the System for Atmospheric Modeling—Explicit Microphysics (SAMEX).
The dynamical core is based on the System for Atmospheric Modeling (SAM; Khairoutdinov and Randall
2003), a model based on an anelastic equation set and
monotonic, positive-definite advection for scalar quantities (Smolarkiewicz and Grabowski 1990). SAMEX is
applied in the near-LES mode described above. For
these simulations, SAMEX employs a horizontal grid
spacing of 150 m. The vertical mesh is stretched, with
grid spacing ranging from 25 m near the surface and the
inversion, with larger values between. At these grid
spacings, the near-LES runs with domain size 57.6 3
57.6 km2 capture the mesoscale variability reasonably
well. This experimental setup is employed as a starting
point for the simulation suite.
Previous studies (Bretherton et al. 1999; Stevens et al.
2005a) emphasized the importance of resolving the entrainment zone, and the fact that the vertical grid spacing
should be O(5 m) in order to represent entrainment correctly. Because our simulations are outside the realm of
true LES, we are particularly concerned about adequately
representing the entrainment process. The model subgrid-scale (SGS) parameterization is that of Deardorff
(1980), with the SGS grid length taken to be the local
vertical grid spacing, as in Khairoutdinov and Randall
(2003). While we acknowledge the limitations of 25-m
vertical grid spacing, we have found that indirect measures of entrainment rates, such as being able to predict
reasonable cloud LWP values and the behavior of the
smaller-scale cloud processes in our simulations, compare well to a traditional high-resolution LES of the
same case.
SAMEX employs explicit microphysics (Kogan 1991),
which represent both droplets and CCN in a size-resolving
framework. Simulations were conducted with 34 droplet
bins and 19 CCN bins. The number of bins is based on
1254
JOURNAL OF THE ATMOSPHERIC SCIENCES
VOLUME 69
FIG. 4. Time–height evolution of (top) potential temperature u and (bottom) water vapor
mixing ratio for 26–28 Oct 2008. The arrow below the x axis of the bottom panel indicates the
time of the sounding used to initialize the simulations.
simulations from the Rain in Cumulus over the Ocean
(RICO) trade cumulus intercomparison (van Zanten et al.
2011). The strong drizzle cases in VOCALS benefit from
the relatively large number of droplet bins (34), compared
to previous simulations of stratocumulus, which employed
25 bins. (Kogan et al. 1995; Khairoutdinov and Kogan
1999). Although computationally expensive, bin microphysics assures the best possible fidelity for aerosol–
cloud–precipitation interactions, as well as for calculations
such as reflectivity.
moister boundary layers, which we suspect arise from
evaporating drizzle. The boundary layer during the period of interest is anomalously deep (zi ’ 1650 m) and
moist, with a surface qy of greater than 10 g kg21.
Model initial conditions are based on the 1100 UTC 26
October 2008 sounding from the RHB. This sounding
b. Initial conditions and forcing
Our simulations are centered around the heavy precipitation features present in the RHB C-band radar from
1100 to 1300 UTC 26 Oct 2008 (Fig. 1). Soundings from
the RHB taken over a 3-day period indicate a boundary
layer with mean depth zi of 1440 m that ranges in depth
from 1185 to 1631 m (Fig. 4). Although the boundary layer
depth varies, the stability of the inversion layer changes
little over the course of the 3 days. Free-tropospheric
humidity is very low, less than 1 g kg21. Periods of large
zi tend to be associated with cooler and sometimes
FIG. 5. Idealized soundings overlaid on the 1100 UTC 26 Oct 2008
RHB sounding.
APRIL 2012
MECHEM ET AL.
(Fig. 5) indicates a boundary layer approximately 1650 m
deep with a potential temperature jump Du 5 12.7 K and
a moisture jump Dqy 5 28.25 g kg21. In contrast to
typical stratocumulus cases with well-mixed boundary
layers, this VOCALS sounding is highly stratified. Because this sounding is associated with periods of large
precipitation rates, this stratification may partly be a result of the precipitation process acting to stably stratify
the boundary layer (Stevens et al. 1998). The model initial
profiles are constructed as a simple piecewise-linear fit to
the observed profiles (Fig. 5). This piecewise-linear fit is
responsible for the slight difference in boundary layer
depth between the sounding (1631 m) and that used in the
simulations (1650 m).
Simulations assume only longwave radiation is active
(nocturnal conditions), and for reasons of computational
expense we employ the simple longwave treatment based
on liquid water path previously employed in Stevens et al.
(2005a) and Ackerman et al. (2009). Although this treatment of radiation is highly idealized, it nevertheless is
remarkably accurate, assuming it is calibrated for the
specific case (Larson et al. 2007). Although the simple
approach neglects clear-sky radiative cooling, which has
been found to be important in boundary layer open cells
(Wang and Feingold 2009a,b), the high cloud fraction of
this case (.0.89) justifies its use here.
Parameters for the radiation scheme were evaluated by
applying the d–four-stream radiative transfer method of
Fu and Liou (1992, 1993) to the 1100 UTC sounding. In
addition to the sounding, we assumed a simple adiabatic
liquid water profile with a maximum LWC of 0.5 g m23 at
a height of 1.65 km. Calculations of net radiative flux
across the boundary layer are for the most part insensitive
to this rather arbitrary choice of maximum LWC. These
calculations resulted in an estimate of 105 W m22 of
cooling across cloud top and 25 W m22 of warming across
cloud base, or a net flux of 80 W m22 across the boundary
layer depth. This large radiative flux reflects the dryness
of the free troposphere and is one distinguishing characteristic of boundary layer cloudiness over the VOCALS
region, a result consistent with aircraft observations of
91 W m22 in overcast regions (Wood et al. 2011). We
projected the cloud-base and cloud-top fluxes onto a single value of 80 W m22 applied at cloud top. Stevens et al.
(2005a) argued that simulations employing this simplification differed little from those employing separately the
cloud-top cooling and cloud-base warming terms.
The bin microphysical parameterization requires initial
conditions for CCN. We assumed a CCN concentration
of 135 cm23, which we obtained from an average of RHB
CCN measurements (0.6% supersaturation) over the
0700–1700 UTC period (D. Covert 1999, personal communication). Without having detailed size information,
1255
we assume a CCN size distribution follows the same
shape as the lognormal spectrum used in the RICO intercomparison (van Zanten et al. 2011). Future simulations may take advantage of CCN activation spectra
calculated from concentration measurements taken at
various magnitudes of supersaturation. The CCN particles were assumed to be completely soluble and composed of sodium chloride. Although it is likely the
observed CCN were some combination of sodium chloride and ammonium sulfate, the model is not particularly
sensitive to this assumption of composition.
Surface fluxes of heat, moisture, and momentum are
evaluated via a bulk aerodynamic framework and calculated using the ocean-relative winds. Sea surface
temperature (SST) is assumed to be 291.4 K. These assumptions result in fluxes of 4–8 W m22 for sensible
heat and 55–70 W m22 for latent heat. The latent heat
flux produced in the simulations is somewhat smaller
than both the observed value of 95 W m22 and the climatological estimate of 60–100 W m22 calculated from
a synthesis of field campaigns over the SEP (de Szoeke
et al. 2010a). We suspect the evaporation of strong
precipitation in the simulation moistens the surface
layer and reduces the moisture flux, relative to the SEP
climatology, but the discrepancy between the model and
surface latent heat flux for this particular case is puzzling. Strong evaporation leading to an underestimate of
latent heat flux would also produce an overestimation of
sensible heat flux. However, in our simulations sensible
heat flux is underestimated as well (observed sensible
heat flux value of 15 W m22). The underestimates in
the fluxes are consistent with a slight underestimate in
surface winds produced by the model (5.3 m s21 in the
model vs 7.8 m s21 in the observations).
A time-independent profile of large-scale vertical
motion (subsidence divergence) is imposed throughout
the simulation. Vertical velocity wls is zero at the surface and decreases linearly to a value of 20.165 cm s21
at an altitude of 1.65 km and above. This vertical velocity profile was derived from the 850-mb vertical
motion field in National Centers for Environmental
Prediction (NCEP) analysis, averaged over the course
of the day and over a domain ranging from 878 to 838W
in longitude and from 228 to 188S in latitude. Although
the location in the vicinity of 858W experiences noticeable day-to-day variability in subsidence, its remote
location relative to the continent minimizes the influence of the upsidence wave (Garreaud and Muñoz
2004; Wood et al. 2009).
c. Sensitivity experiments
A significant advantage of a numerical modeling approach is the ability to isolate the effects of specific factors
1256
JOURNAL OF THE ATMOSPHERIC SCIENCES
VOLUME 69
FIG. 6. PDFs of the inversion height zi and CCN concentration (S 5 0.6%) The ‘‘3’’ symbols represent the inversion
height and CCN concentration of the Deep (control) simulation.
or combinations of factors via sensitivity experiments,
where different factors are varied from a control simulation. We test our hypothesis (‘‘Natural variability
in boundary-layer thermodynamic properties exerts a
greater influence in cloud and precipitation processes
than does variability in aerosol concentrations.’’) using
this approach, isolating the effects of boundary layer
depth and CCN concentration. We vary these two quantities based on the variability of these factors observed
during the VOCALS cruises. Constraining these quantities according to natural variability serves to assess the
relative importance of the two factors.
From the point of view of variability over the course of
the entire VOCALS project, boundary layer depth and
CCN concentrations at the 1100 UTC sounding time are
deeper and cleaner than average (Fig. 6). In one simulation (termed ‘‘Shallow’’), we decrease the boundary
layer depth by 200 m, corresponding approximately to
a 1s decrease in the inversion height probability distribution function (PDF) (Fig. 6). In a similar fashion, we
perform a simulation (‘‘Doubled CCN’’) in which CCN
concentration is doubled from 135 to 270 cm23, again
corresponding approximately to a 1s increase.
Owing to the likelihood of mutual interactions between the two factors, sensitivity experiments varying
both factors are performed. From a dynamical systems
standpoint, this is an important additional step, since it is
unknown whether these combined interactions are
simply additive in nature, whether they tend to cancel,
or whether variations in multiple factors result in interactions that are nonlinear and unpredictable. Following the factor separation method described in Stein
and Alpert (1993) and Dearden (2009), we perform an
additional simulation (‘‘Shallow 1 Doubled CCN’’),
which takes into account the mutual interactions between these two factors.
3. Results
All four simulations were run for 18 h. In this paper, we
concentrate our analysis on the period from 8 to 12 h in
order to focus on what we term the primary microphysical–
dynamical responses. Time series over longer periods give
evidence of complicated feedbacks, which are beyond the
scope of our study. To spin up boundary layer turbulent
flow, collision–coalescence was turned off for the first 6 h
of the simulation.
Over the period from 8–12 h, the cloud system in the
control simulation begins as an unbroken, relatively homogeneous cloud deck and develops a substantial degree
of mesoscale structure, characterized by large drizzle cells
of order 100 km2. Figure 7 shows an example of these
drizzle cells, indicated by the maximum reflectivity in
each column (composite reflectivity). Reflectivity is calculated using the entire drop size distribution available
from the model.
A vertical cross section through one of these mesoscale
drizzle cells (Fig. 7b) is reminiscent of organized deep
convective structures (e.g., Zipser 1977; Houze et al.
1989). The region of strongest precipitation in the cell (at
X 5 28 km) is associated with a tilted updraft structure
ranging from the surface to the top of the boundary layer,
and downdraft in the precipitation core itself at levels
below 0.6 km. The regions of weaker precipitation to the
east are also downdraft-dominated. Despite continuous
radiative forcing in the form of cloud-top cooling, vertical
motion in the overhanging ‘‘anvil’’ cloud, at least in this
particular cross section, is weak. The simulated radar image suggests that strong, penetrative updrafts associated
with individual drizzle cells are evidently acting as a substantial moisture source for the stratocumulus cloud layer.
In fact, it appears that liquid water detraining from these
strong but relatively isolated updrafts is the predominant
APRIL 2012
MECHEM ET AL.
1257
FIG. 8. Time series of surface precipitation rate R, liquid water
path, cloud fraction, and droplet concentration.
FIG. 7. Model reflectivity at 12 h for the control simulation.
(a) Composite reflectivity. Thick black contours are through 0 dBZ
points and roughly represent the sensitivity of the RHB C-band
radar. (b),(c) Vertical cross sections of reflectivity and vertical
velocity through the red lines in (a). Contours represent vertical
velocity values of 22.5, 20.5, 0.5, and 2.5 m s21, with dashed
contours representing negative values.
factor maintaining the stratocumulus cloud deck. These
cloud structures bear significant resemblance to those in
the Atlantic Trade Wind Experiment (ATEX) model intercomparison (Stevens et al. 2001), in which trade cumulus detraining at the inversion serves to maintain the
stratocumulus layer.
a. Mean simulation behavior
Time series over a 4-h period indicate marked differences among the four simulations (Fig. 8). Precipitation
rates R are highly variable, with the Deep (control)
simulation having the largest rate and the Shallow 1
Doubled CCN simulation having the smallest. Although
the time series are noisy, R for the Doubled CCN simulation is larger than for the Shallow simulation, indicating that a reduction in boundary layer depth exerts
more of an impact on precipitation outcomes than does
a typically observed increase in CCN. Early in the simulation (8–9 h), precipitation rate and LWP in the
Shallow simulation are larger than in the Deep case. We
speculate this behavior is associated with the slightly
greater precipitation produced early in the Deep simulation and the subsequent stabilization of the boundary
layer, which serves as a brake on turbulent kinetic energy (TKE) production.
LWP slightly leads the precipitation rate, which implicates LWP as a primary driver in the precipitation process.
In each simulation, the LWP increases for a time, followed
by a leveling off or a decrease that appears to coincide with
a precipitation rate threshold, different for each simulation. The temporal behavior of LWC is complicated,
however, because it not only acts as a control on drizzle
production but also is strongly affected by drizzle (via
sedimentation and dynamical feedback on the entrainment rate). Cloud fraction fc is smallest in the most strongly
precipitating (Deep) case but nevertheless remains high
(;0.89). In the other four simulations, fc remains above
0.96, indicating a persistent deck of stratocumulus. Droplet
concentration Nc is initially determined by the CCN concentration. As would be expected, the two simulations with
fewer CCN precipitate more, and hence the fractional reduction of Nc with time via coalescence processing is
greater (a positive feedback between precipitation rate
and Nc reduction). Coalescence processing has been shown
to be dependent on both precipitation rate and on the
1258
JOURNAL OF THE ATMOSPHERIC SCIENCES
VOLUME 69
TABLE 1. Time mean (median in parentheses) of LES precipitation rate, LWP, and cloud fraction, averaged over the 8–12-h
period. Entrainment rate is calculated as we 5 dzi/dt 2 wls, where zi
is the inversion height and wls is the specified large-scale vertical
motion (m s21) at the inversion.
Simulation
Deep
Shallow
Doubled CCN
Shallow 1
Doubled CCN
R
(mm day21)
LWP
(g m22)
fc
we
(cm s21)
0.98 (1.15)
0.44 (0.48)
0.57 (0.45)
0.28 (0.31)
145 (148)
125 (128)
151 (163)
139 (149)
0.96 (0.96)
0.97 (0.96)
0.99 (0.99)
0.99 (0.99)
0.76
0.58
0.85
0.72
FIG. 9. Time series of accumulated precipitation.
droplet concentration itself (Mechem et al. 2006; Wood
2006).
Table 1 summarizes the time-mean behavior of these
statistics. The precipitation rate is strongest for the Deep
simulation. Reducing the boundary layer depth by
200 m results in 0.54 mm day21 decrease (255%) in R.
Doubling the CCN reduces R but not quite so dramatically (20.41 mm day21 or 242%). Differences in mean
LWP between the simulations are not quite as dramatic.
Liquid water path is reduced in the Shallow simulation
by 14% and in the Doubled CCN simulation by 4%.
Furthermore, differences in LWP between the simulations may partly be an artifact of simply averaging over
the last 4 h of the simulations, rather than choosing caseby-case averaging periods defined by the evolution of
the cloud and precipitation fields. Even in the most
strongly precipitating case (Deep simulation), the mean
cloud fraction remains large, although the time series in
Fig. 8 indicates that cloud fraction in the Deep simulation is departing from the ensemble and becoming notably smaller. From what is effectively just a single
statistical realization, we are not able to evaluate the
statistical significance of these differences.
Entrainment rate we is generally consistent with LWP
and precipitation rate (Table 1). This relationship is
clearest between the two deep simulations (Deep and
Doubled CCN) and between the two shallow simulations
(Shallow and Shallow 1 Doubled CCN). The shallow
versus deep simulations have large differences in their
baseline thermodynamic and cloud structures. In each set
of simulations (i.e., comparing Deep with Doubled CCN,
or comparing Shallow with Shallow 1 Doubled CCN),
increased precipitation is associated with decreased liquid water path, reduced cloud fraction, and a smaller
entrainment rate. This behavior is consistent with drizzle
reducing cloud water (either by sedimentation or by
suppressing the vertical transport via stably stratifying the
boundary layer), leading to a reduction of radiative
cooling at cloud top. Less radiative cooling and less liquid
water at cloud top (reducing evaporative enhancement
associated with entrainment mixing of free tropospheric
and cloudy boundary layer air) can both act to reduce
TKE and entrainment. One caveat applies to our approach, and to others that use this near-LES methodology:
given our relatively crude approximately 25-m vertical
grid spacing near the inversion, we note that we is most
likely overestimated. Entrainment rates during a 6-day
period from the EPIC field campaign were estimated to be
0.4 6 0.1 cm s21 (Caldwell et al. 2005), although we would
expect we from our nocturnal simulations to perhaps be
somewhat larger than the EPIC estimates.
Although precipitation rate most directly corresponds
to other boundary layer energetic fluxes (i.e., it can be
expressed as watts per square meter), because of its
noisiness the accumulated precipitation is often a better
statistic to gauge the resulting precipitation behavior.
Accumulated precipitation can be plotted as a time series, with the slope of the time series corresponding to the
precipitation rate (Fig. 9). This presentation confirms that
the Deep case precipitates most strongly, followed by the
Doubled CCN, the Shallow, and the simultaneously
Shallow 1 Doubled CCN cases. Separating the cases as
above into deep and shallow simulations indicates that
the main effect of additional CCN is to delay the onset of
precipitation (as in Stevens and Seifert 2008) but that
after a time the precipitation rates for the clean and the
doubled CCN simulations are roughly similar. This behavior most likely stems from the increased CCN suppressing the self-collection of small drops into precipitation
nuclei. Once a sufficient number of precipitation nuclei are
created, however, collection of cloud droplets by these
precipitation nuclei can occur just as it does in the cleaner
simulation. Profiles of mean precipitation rate (Fig. 10) are
consistent with the surface precipitation rate time series
but also indicate that precipitation evaporation in the
subcloud layer plays a substantial role in influencing surface
precipitation. In particular, the low-CCN-concentration
simulations experience greater evaporation than simulations with greater CCN concentration.
APRIL 2012
MECHEM ET AL.
FIG. 10. Mean profiles of precipitation rate. The plus signs represent
mean cloud base and cloud top.
The boundary layer is greatly stratified thermodynamically, with a difference in ul of about 1.5 K between
the surface and the inversion base (Fig. 11). Moisture
is even more stratified, with qt decreasing by about
2.5 g kg21 between the surface and inversion base.
Mean cloud base is approximately 450 m and delineates
distinct cloud-layer and subcloud-layer structures. The
bulk of the moisture stratification takes place between
450 and 800 m in the lower part of the cloud layer. The
bulk of the cloud layer from 800 m up to the inversion is
reasonably well mixed. The liquid water profiles (cloud
water plus precipitation) exhibit peaks in the upper part
of the cloud (;1.65 km in the Deep simulation), but the
peak is not sharp and the profile is far from adiabatic
(near linear) with height. The profiles in Fig. 11 suggest
1259
substantial differences between this case and idealized,
nonprecipitating stratocumulus. In idealized stratocumulus, ul and qt are more or less well mixed over the
entire boundary layer, and average liquid water ql exhibits a sharp maximum at the inversion base, with a linear decrease below the maximum. Strong precipitation
acts to maintain the thermodynamic stratification in Fig.
11 via evaporation of drizzle over the subcloud layer and
the resulting net cloud-layer warming proportional to the
precipitation rate (Stevens et al. 1998).
Except for the obvious difference in boundary layer
depth, visually the ul profiles do not differ substantially
between the four cases, but calculations confirm that
the Deep (control) is the most stable of the profiles
(dul/dz 5 1.65 K km21 for the Deep simulation compared
to 1.07 K km21 for the Shallow 1 Doubled CCN simulation). The stabilizing effects of precipitation are more
evident in the qt profiles, with the deeper, more strongly
precipitating simulations characterized by a greater difference in qt between inversion base and the surface.
Average liquid water in the cloud layer is most strongly
affected by changes in CCN, suggesting that CCN may
have more of an impact than PBL depth on cloud optical
properties.
Profiles of dynamical quantities (Fig. 12) describe
boundary layers with the greatest vertical velocity variance w9w9 (i.e., strongest turbulence) lying in the cloud
layer. Although the qt profiles in Fig. 11 display evidence
of two distinct layers, the vertical velocity variance does
not exhibit the double-maximum structure typical of two
distinct circulations found in decoupled, radiatively
driven stratocumulus.
The two deeper simulations have greater maxima in
w9w9. For a given boundary layer depth (e.g., comparing
the Deep and Doubled CCN simulations), the maximum
in w9w9 is well correlated with LWP, which is evident
comparing the w9w9 profiles in Fig. 12 with the ql profiles
FIG. 11. Mean profiles of liquid water potential temperature ul, total water mixing ratio qt, and liquid water mixing ratio ql.
1260
JOURNAL OF THE ATMOSPHERIC SCIENCES
VOLUME 69
3/2
FIG. 12. Mean profiles of vertical velocity variance w9w9, vertical velocity skewness w9w9w9/w9w9 , and buoyancy flux rcp w9u9y . The plus
signs represent mean cloud base and cloud top.
in Fig. 11. Boundary layer depth also influences the w9w9
maximum through the vertical distance over which the
buoyancy generation mechanism acts: all else being
equal, a deeper boundary layer will exhibit a greater
maximum in w9w9.
Profiles of skewness (Fig. 12) are positive everywhere,
consistent with smaller, more intense updrafts, and
larger, weaker downdrafts. Positive skewness implies a
circulation driven in a bottom-up fashion, predominantly by convection rooted in the surface layer. The
tendency of the vertical velocity field to be positively
skewed in strongly precipitating stratocumulus-topped
boundary layers is well known (e.g., Stevens et al. 1998;
Ackerman et al. 2009), a behavior that is also evident in
trade cumulus boundary layers (Siebesma et al. 2003;
van Zanten et al. 2011). All four simulations are positively skewed, although there is a tendency for the
skewness in the deeper cases to be greater.
The buoyancy flux is predominantly positive over the
depth of the boundary layer, except for the layer of negative flux near the inversion and a very shallow layer in the
doubled CCN simulation at z 5 0.5 km. Previous work
showed that, most often, the decoupling of the cloud and
subcloud properties evinced in the mean profiles is accompanied by negative buoyancy fluxes and a minimum
in vertical velocity variance at cloud base (e.g., Ackerman
et al. 2009). The VOCALS simulations, on the other hand,
exhibit no cloud-base minimum in w9w9, and the buoyancy flux is (nearly) uniformly positive. The main difference between simulations is the magnitude of positive
buoyancy flux. The maxima of buoyancy flux are greater
in the deeper cases, meaning that the buoyancy generation of TKE is greater in these simulations. From the
positive skewness and buoyancy flux we may infer that the
updrafts are, at least in the mean sense, always positively
buoyant and never have to do work against the mean
stratification. That is, the circulation is always associated
with a transfer of potential energy (associated with parcels based in the subcloud layer) into kinetic energy.
Similarly, downdrafts are negatively buoyant on average,
although the positive skewness indicates that the circulation is updraft-dominant. A comparison of Table 1 and
Fig. 12 illustrates that entrainment rate increases with
increasing integrated buoyancy flux (as in Caldwell and
Bretherton 2009), which gives us some confidence that the
model’s entrainment rate behavior is reasonable (even
though the entrainment rates themselves are too large).
b. Statistical distributions
The C-band radar on the RHB can provide constraints
on the control (Deep) simulation. Figure 13 compares
radar reflectivity PDFs of the four simulations, with the
FIG. 13. Normalized frequency distribution (PDF) of reflectivity
calculated at a constant height of 490 m, taken over the 4-h period
(8–12 h). Units are fraction per dBZ.
APRIL 2012
MECHEM ET AL.
1261
FIG. 14. Cumulative and normalized frequency distribution (PDF) of surface precipitation rate, calculated over the
last 4 h of the simulations.
reflectivity PDF from the RHB from (corresponding to
the strong drizzle case from 0900 to 1700 UTC) overlaid.
The figure compares reflectivity PDFs taken at a constant
height of 490 m. In the model, reflectivity is calculated as
the sixth moment of the drop size distribution over the
4-h period (8–12 h). The observational PDF is calculated
over the period 0900–1700 UTC, over which the linear
feature in Fig. 1 is present. A reflectivity threshold of
210 dBZ is applied prior to calculating the PDF from the
simulation output. This threshold roughly corresponds
to the sensitivity of the C-band radar in the immediate
vicinity of the ship.
Comparing model-calculated reflectivity with observations is fraught with challenges. Most fundamentally,
the model-derived reflectivity is a point calculation,
which neglects subgrid-scale variability in drop spectra
that is captured by the volume sampling of the radar. The
radar sampling volume increases with range, whereas the
model grid size is consistent over the domain (except for
the stretched vertical grid). The caveats about comparing
observed and simulated reflectivity notwithstanding, the
simulated reflectivity PDFs agree surprisingly well with
the PDF calculated from observed reflectivity (Fig. 13).
Ideally, the Deep (control) simulation would exhibit the
best agreement with the observations, but given the idealized nature of the simulations and the difficulties with
comparing observed and model-derived reflectivities, the
agreement between simulated reflectivity and the observations is remarkable. All four simulations exhibit similar
distributions for reflectivities up to approximately 28 dBZ.
Above this threshold the distributions diverge. The simulations, particularly the shallow cases, capture the observed occurrence of reflectivity greater than 30 dBZ.
PDFs of surface precipitation rate (Fig. 14) indicate
contributions of showery precipitation consistent with the
reflectivity PDFs. The precipitation PDFs exhibit lognormal behavior, with the modal values ranging from 0.3
to 1.0 mm day21 consistent with the mean R values in
Table 1. The cumulative distributions further illustrate
the relative contribution from large precipitation rates.
For the control simulation, precipitation rates greater
than 1 mm day21 contribute 43% of the precipitation, and
rates greater than 10 mm day21 contribute 15%. Rates
greater than 100 mm day21 contribute a small but nonnegligible fraction (2%) of the total precipitation.
To more completely characterize the statistical distribution of the model reflectivity, we calculate contoured
frequency by altitude diagrams (CFADs; Yuter and Houze
1995) for the four simulations (Fig. 15). The CFADs include the entire range of simulated reflectivity from 240 to
50 dBZ, ranging well below the sensitivity of the RHB radar. CFADs from all four cases are superficially quite similar. The consistency of the modal reflectivity (;210 dBZ
in all the cases) is consistent with the similarity in the
liquid water content and other thermodynamic profiles in
Fig. 11. The increase with height of the modal reflectivity
indicated by the dashed line on the Deep simulation panel
corresponds to the increase in liquid water content (mostly
via droplet size) with height in updrafts. Because many of
these updraft structures are reminiscent of trade cumulus
updrafts, we anticipate that their liquid water content
values are substantially subadiabatic. The reflectivity mode
associated with the precipitation core is evident below
0.6 km. For a given drop spectrum, smaller drops are
preferentially evaporated as they fall and large drops
continue to coagulate, which serves to explain the increases in modal reflectivity as precipitation falls toward
the surface, even though the total precipitation decreases.
The most striking difference between the simulations
relates to the instances of large values of reflectivity in the
distribution tail at each altitude. The reflectivity corresponding to the smallest contour value (1025) gives a
rough indication of the frequency of occurrence of large
reflectivities in each simulation. The largest reflectivity
values are slightly less frequent in the Doubled CCN
simulation but notably reduced in the two shallow cases.
1262
JOURNAL OF THE ATMOSPHERIC SCIENCES
VOLUME 69
FIG. 15. CFADs of reflectivity for the four simulations. Labels on the color bar correspond to contour values. Contour
units are fraction per dBZ.
A major result of this study is that a perturbation to the
boundary layer thermodynamics (a 1s or 200-m reduction in depth) has a greater influence on reducing the
precipitation than does a 1s change (corresponding in
this case to a doubling) in CCN concentration.
CFADs of the kinematic fields in Fig. 16 are conditionally sampled to include only regions of reflectivity of
at least 10 dBZ, corresponding to drizzle cells cores, to
better compare with the radar-based statistics of SEP
drizzling clouds from Comstock et al. (2007). The mode
FIG. 16. CFADs of (a) divergence and (b) vertical velocity for the Deep simulation, conditionally sampled in regions of reflectivity $
10 dBZ. Thick dashed lines represent mean profiles. (c) Profile of the fractional number of points in each level employed in the CFADs,
related to the number of points in each level. The difference in topmost height of the CFADs is a result of the choice of contour levels.
APRIL 2012
MECHEM ET AL.
of the divergence distribution in Fig. 16a slants to the left,
with the mean profile indicating that the tendency is for
convergence to increase with height within the 0.4–1.1-km
layer. The distribution mode is most strongly convergent
at 1.1 km, just below the altitude of the strongest convergence outliers at 1.2 km. Divergence outliers of similar
magnitude occur near the top and the bottom of the
boundary layer. The detrainment near the inversion is
primarily occurring in echoes less than 10 dBZ, which is
not present in the distribution. Near the surface, the distribution mode is slightly divergent, and the strongest
divergence is about twice as large as the maximum convergence. This structure is consistent with open-cellular
category described in Comstock et al. (2007). In particular, the midlevel convergence is a prominent feature of
both the model and observations. In this simulation the
cloud and subcloud layers are coupled by deep convective
circulations, and the wide layer of midlevel convergence
is associated with surface-based updrafts ascending into
cloud.
The distribution of vertical velocity is by definition
consistent with the horizontal divergence but aids in illustrating details of the updraft/downdraft structure.
Strong updrafts (.4 m s21) occur at 1.4 km in altitude,
whereas downdrafts are present but tend to be weaker
than updrafts (.22 m s21). The thresholding at 10 dBZ
yields modal vertical velocities that are slightly negative.
We do not expect to see evaporatively driven cold pools
in this distribution because their associated reflectivities
are less than 10 dBZ. The positive skewness of the vertical velocity distribution implies the prevalence of updrafts over downdrafts.
4. Discussion and conclusions
We performed a series of simulations based on observations of strongly drizzling boundary layer cells observed during the VOCALS field campaign in order to
address our motivating question. Specifically, we asked
which exerts greater control on boundary layer precipitation outcomes, thermodynamic properties (specifically
in the form of boundary layer depth) or aerosol. The
control case was based on field observations taken from
the RHB. The sensitivity simulations were constrained by
the variability in boundary layer depth and CCN concentration observed over the course of the field project.
Both thermodynamic and CCN perturbations influence precipitation processes, but our findings indicate
that the sensitivity to thermodynamic changes dominates,
at least for the short-term sensitivity explored here. The
deeper boundary layer produces greater liquid water
content and more precipitation. Heavier precipitation
tends to stabilize the boundary layer to a greater degree
1263
than lighter precipitation. Increased aerosol concentration primarily affects the timing of precipitation rather
than precipitation rate. Longer-term feedbacks may be
different because boundary layer cloud feedbacks may
often be damped or buffered (Stevens and Feingold
2009).
Our conclusions about the relative sensitivity of the two
perturbations are similar to those reached by Wang et al.
(2010). Although both studies are based on VOCALS, we
assume environmental characteristics from farther west in
the region when the RHB was located at approximately
858W. Thermodynamically, the boundary layer we simulated was deeper than the Wang et al. case (centered at
808W), contained anomalously stronger precipitation
cells, and was stratified to a greater degree. The two
studies thus explore different regions of the SEP parameter space.
In addition to addressing the primary question, our
results speak to dynamical and precipitation processes in
a highly stably stratified boundary layer. Negative buoyancy arising from cloud-top longwave cooling doubtless
contributes to the buoyancy generation of TKE; however,
we find that the boundary layer circulation is dominated
by a relatively few number of strong couplets composed
of buoyant updrafts and precipitation-laden downdrafts.
The strongest local precipitation rates and the largest
reflectivity values are associated with these cells.
The term ‘‘decoupled’’ might apply to our simulated
boundary layer, yet this term is fraught with much ambiguity. When is a boundary layer decoupled? Is it sufficient that the mean thermodynamic properties of the
cloud and subcloud layers differ? Must a minimum in
cloud-base w9w9 or negative buoyancy flux be present? Is
either condition sufficient to classify a boundary layer as
decoupled, or are both conditions necessary? The large
degree of stratification in our VOCALS simulation is
consistent with the view of decoupling defined by thermodynamic stratification but does not satisfy the dynamical requirements. Yet it is clear that the two layers
are coupled, perhaps intermittently, in the sense that
strong convective updrafts provide the source of moisture
to the cloud layer. Hence, even if the overturning of the
boundary layer is insufficient to mix out the stable stratification, the subcloud and cloud layers nevertheless
communicate via convective updrafts and downdrafts.
The VOCALS boundary layer explored in this study
sits in a unique position in the continuum of marine
boundary layers (Fig. 17). We show two versions of the
profile for VOCALS, one representing the period of
strong drizzle on 26 October 2008 and the other scaling
the 26 October sounding by the mean boundary layer
depth observed by the RHB during the VOCALS cruises.
On one side of the continuum, DYCOMS-II RF01 is an
1264
JOURNAL OF THE ATMOSPHERIC SCIENCES
VOLUME 69
The conceptual model of Wyant et al. (1997) ascribes
this transition predominantly to a deepening and decoupling of the boundary layer stemming from moving
over warmer SSTs. But many modeling studies (e.g.,
Stevens et al. 1998) have demonstrated that precipitation can also result in thermodynamic stratification
and decoupling. The relative importance of precipitation and SST gradient mechanisms in producing the
continuum in Fig. 17 is not well understood and would
be a fruitful topic for further pursuit.
FIG. 17. Conceptual model of the continuum of cloudy boundary
layers. From right to left: boundary layer stratocumulus (DYCOMS
II); the VOCALS environment at 208S, 858W; trade cumulus rising
into a stratocumulus layer (ATEX); and trade cumulus (BOMEX).
The dashed lines in the VOCALS profiles represent the strong
drizzle case of 26 Oct 2008 that we simulate. The solid lines represent
the 26 Oct sounding scaled by the mean boundary layer depth observed during VOCALS.
archetypical marine boundary layer topped by stratocumulus (Stevens et al. 2005a). The DYCOMS-II boundary
layer is characterized by fairly shallow, nearly well-mixed
boundary layers with large jumps at cloud top. On the other
end lie the trade cumulus cases of the Barbados Oceanographic and Meteorological Experiment (BOMEX;
Siebesma et al. 2003) or RICO (van Zanten et al. 2011).
Trade cumulus boundary layers contain a well-mixed
subcloud layer and a thermodynamically stratified cloud
layer topped by the trade inversion. Between these two
extremes of the continuum lies the regime where trade
cumulus clouds rise into a deck of stratocumulus. ATEX
(Stevens et al. 2001) provides a good example of the last
type of cloud system, which contains some characteristics
both of stratocumulus and trade cumulus. More stable
and deeper boundary layers populate the right side of the
continuum; more well-mixed and shallower boundary
layers lie to the left. The vertical structure of the VOCALS
boundary layer, particularly the variation of thermodynamic structure in the vertical, bears greatest resemblance
to ATEX. VOCALS is cooler and drier than ATEX and is
topped by a sharper inversion. In a geographical sense, the
continuum represents an idealized transition from eastern
subtropical oceans (DYCOMS-II) toward the west and
equatorward (BOMEX).
Acknowledgments. Thanks to Marat Kharioutdinov
for making available the SAM model to the scientific
community and to Yefim Kogan for ongoing advice and
collaborations. Dave Covert supplied the CCN data.
Thanks to Qiang Fu for the use of his radiation code.
Thanks to Rob Wood for the NCEP analysis data.
Special thanks go to Chris Fairall, Dan Wolfe, Sergio
Pezoa, Matthew Miller, and Justin Crouch for running
and monitoring C-band radar operations during the
VOCALS cruises and to Christina Aldereguia, Angel
Adames, Monica Laureno, and Casey Burleyson for
radar data processing and analysis. Beth Tully refined
some of the figures. Some of the computing for this
project was performed at the OU Supercomputing
Center for Education and Research (OSCER) at the
University of Oklahoma. We appreciate the thoughtful
comments from three anonymous reviewers. This investigation was supported by the University of Kansas
General Research Fund allocation 230211 and New
Faculty Startup funds, and U.S. National Oceanic and
Atmospheric Administration (NOAA) Climate Program Office (CPO) Climate Prediction Program for the
Americas (CPPA) Grants GC09-252b, GC09-507, and
NA10OAR4310160.
REFERENCES
Ackerman, A. S., and Coauthors, 2009: Large-eddy simulations of
a drizzling, stratocumulus-topped marine boundary layer.
Mon. Wea. Rev., 137, 1083–1110.
Albrecht, B. A., 1989: Aerosols, cloud microphysics, and fractional
cloudiness. Science, 245, 1227–1230.
Bony, S., and J. Dufresne, 2005: Marine boundary layer clouds
at the heart of tropical cloud feedback uncertainties in climate models. Geophys. Res. Lett., 32, L20806, doi:10.1029/
2005GL023851.
Bretherton, C. S., and Coauthors, 1999: An intercomparison of
radiatively driven entrainment and turbulence in a smoke
cloud, as simulated by different numerical models. Quart.
J. Roy. Meteor. Soc., 125, 391–423.
——, and Coauthors, 2004: The EPIC 2001 stratocumulus study.
Bull. Amer. Meteor. Soc., 85, 967–977.
Caldwell, P., and C. S. Bretherton, 2009: Large eddy simulation of
the diurnal cycle in southeast Pacific stratocumulus. J. Atmos.
Sci., 66, 432–449.
APRIL 2012
MECHEM ET AL.
——, ——, and R. Wood, 2005: Mixed-layer budget analysis of the
diurnal cycle of entrainment in southeast Pacific stratocumulus. J. Atmos. Sci., 62, 3775–3791.
Comstock, K. K., C. S. Bretherton, and S. E. Yuter, 2005: Mesoscale variability and drizzle in southeast Pacific stratocumulus.
J. Atmos. Sci., 62, 3792–3807.
——, S. E. Yuter, R. Wood, and C. S. Bretherton, 2007: The threedimensional structure and kinematics of drizzling stratocumulus. Mon. Wea. Rev., 135, 3767–3784.
Dearden, C., 2009: Investigating the simulation of cloud microphysical processes in numerical models using a one-dimensional
dynamical framework. Atmos. Sci. Lett., 10, 207–214.
Deardorff, J. W., 1980: Stratocumulus-capped mixed layers derived
from a three-dimensional model. Bound.-Layer Meteor., 18,
495–527.
de Szoeke, S. P., C. W. Fairall, D. E. Wolfe, L. Bariteau, and
P. Zuidema, 2010a: Surface flux observations on the southeastern tropical Pacific Ocean and attribution of SST errors in
coupled ocean–atmosphere models. J. Climate, 23, 4152–4174.
——, S. E. Yuter, P. Zuidema, C. W. Fairall, and W. A. Brewer,
2010b: Ship-based observation of drizzling stratocumulus
clouds from EPIC to VOCALS. CLIVAR Exchanges, No. 39,
International CLIVAR Project Office, Southampton, United
Kingdom, 11–13.
Feingold, G., and S. M. Kreidenweis, 2002: Cloud processing of
aerosol as modeled by a large eddy simulation with coupled
microphysics and aqueous chemistry. J. Geophys. Res., 107,
4687, doi:10.1029/2002JD002054.
——, ——, B. Stevens, and W. R. Cotton, 1996: Numerical simulations of stratocumulus processing of cloud condensation
nuclei through collision–coalescence. J. Geophys. Res., 101,
21 391–21 402.
——, W. L. Eberhard, D. E. Veron, and M. Previdi, 2003: First
measurements of the Twomey indirect effect using ground-based
remote sensors. Geophys. Res. Lett., 30, 1287, doi:10.1029/
2002GL016633.
——, R. Furrer, P. Pilewskie, L. A. Remer, Q. Min, and H. Jonsson,
2006: Aerosol indirect effects at Southern Great Plains during
the May 2003 intensive operations period. J. Geophys. Res.,
111, D05S14, doi:10.1029/2004JD005648.
Frisch, A. S., C. W. Fairall, and J. B. Snider, 1995: Measurement of
stratus cloud and drizzle parameters in ASTEX with a Ka-band
Doppler radar and a microwave radiometer. J. Atmos. Sci., 52,
2788–2799.
Fu, Q., and K. N. Liou, 1992: On the correlated k-distribution
method for radiative transfer in nonhomogeneous atmospheres. J. Atmos. Sci., 49, 2139–2156.
——, and ——, 1993: Parameterization of the radiative properties
of cirrus clouds. J. Atmos. Sci., 50, 2008–2025.
Garreaud, R. D., and R. Muñoz, 2004: The diurnal cycle in circulation and cloudiness over the subtropical southeast Pacific: A
modeling study. J. Climate, 17, 1699–1710.
Garrett, T. J., and P. V. Hobbs, 1995: Long-range transport of
continental aerosols over the Atlantic Ocean and their effects
on cloud structures. J. Atmos. Sci., 52, 2977–2984.
Houze, R. A., S. A. Rutledge, M. I. Biggerstaff, and B. F. Smull,
1989: Interpretation of Doppler weather radar displays of
midlatitude mesoscale convective systems. Bull. Amer. Meteor. Soc., 70, 608–619.
Hudson, J. G., 1993: Cloud condensation nuclei near marine cumulus. J. Geophys. Res., 98, 2693–2702.
Khairoutdinov, M. F., and Y. L. Kogan, 1999: A large eddy simulation model with explicit microphysics: Validation against
1265
aircraft observations of a stratocumulus-topped boundary
layer. J. Atmos. Sci., 56, 2115–2131.
——, and D. A. Randall, 2003: Cloud resolving modeling of the
ARM summer 1997 IOP: Model formulation, results, uncertainties, and sensitivities. J. Atmos. Sci., 60, 607–625.
Klein, S. A., and D. L. Hartmann, 1993: The seasonal cycle of low
stratiform clouds. J. Climate, 6, 1587–1606.
Kogan, Y. L., 1991: The simulation of a convective cloud in a 3-D
model with explicit microphysics. Part I: Model description
and sensitivity experiments. J. Atmos. Sci., 48, 1160–1189.
——, M. P. Khairoutdinov, D. K. Lilly, Z. N. Kogan, and Q. Liu,
1995: Modeling of stratocumulus cloud layers in a large eddy
simulation model with explicit microphysics. J. Atmos. Sci., 52,
2923–2940.
Larson, V. E., K. E. Kotenberg, and N. B. Wood, 2007: An analytic
longwave radiation formula for liquid layer clouds. Mon. Wea.
Rev., 135, 689–699.
Mechem, D. B., P. C. Robinson, and Y. L. Kogan, 2006: Processing
of cloud condensation nuclei by collision-coalescence in a
mesoscale model. J. Geophys. Res., 111, D18204, doi:10.1029/
2006JD007183.
Medeiros, B., B. Stevens, I. M. Held, M. Zhao, D. L. Williamson,
J. G. Olson, and C. S. Bretherton, 2008: Aquaplanets, climate
sensitivity, and low clouds. J. Climate, 21, 4974–4991.
Paluch, I. R., and D. H. Lenschow, 1991: Stratiform cloud formation in the marine boundary layer. J. Atmos. Sci., 48, 2141–
2158.
Savic-Jovcic, V., and B. Stevens, 2008: The structure and mesoscale
organization of precipitating stratocumulus. J. Atmos. Sci., 65,
1587–1605.
Siebesma, A. P., and Coauthors, 2003: A large eddy simulation
intercomparison study of shallow cumulus convection. J. Atmos. Sci., 60, 1201–1219.
Smolarkiewicz, P. K., and W. W. Grabowski, 1990: The multidimensional positive definite advection transport algorithm:
Non-oscillatory option. J. Comput. Phys., 86, 355–375.
Stein, U., and P. Alpert, 1993: Factor separation in numerical
simulations. J. Atmos. Sci., 50, 2107–2115.
Stevens, B., and A. Seifert, 2008: Understanding macrophysical
outcomes of microphysical choices in simulations of shallow
cumulus convection. J. Meteor. Soc. Japan, 86A, 143–162.
——, and G. Feingold, 2009: Untangling aerosol effects on clouds
and precipitation in a buffered system. Nature, 461, 607–613,
doi:10.1038/nature08281.
——, W. R. Cotton, G. Feingold, and C.-H. Moeng, 1998: Large-eddy
simulations of strongly precipitating, shallow, stratocumulustopped boundary layers. J. Atmos. Sci., 55, 3616–3638.
——, and Coauthors, 2001: Simulations of trade wind cumuli under
a strong inversion. J. Atmos. Sci., 58, 1870–1891.
——, and Coauthors, 2003: Dynamics and chemistry of marine
stratocumulus—DYCOMS-II. Bull. Amer. Meteor. Soc., 84,
579–593.
——, and Coauthors, 2005a: Evaluation of large-eddy simulations
via observations of nocturnal marine stratocumulus. Mon.
Wea. Rev., 133, 1443–1462.
——, G. Vali, K. Comstock, R. Wood, M. C. van Zanten, P. H.
Austin, C. S. Bretherton, and D. H. Lenschow, 2005b: Pockets
of open cells and drizzle in marine stratocumulus. Bull. Amer.
Meteor. Soc., 86, 51–57.
Twomey, S., 1974: Pollution and the planetary albedo. Atmos.
Environ., 8, 1251–1256.
——, 1977: The influence of pollution on the shortwave albedo of
clouds. J. Atmos. Sci., 34, 1149–1152.
1266
JOURNAL OF THE ATMOSPHERIC SCIENCES
van Zanten, M. C., and Coauthors, 2011: Controls on precipitation
and cloudiness in simulations of trade-wind cumulus as observed during RICO. J. Adv. Model. Earth Syst., 3, M06001,
doi:10.1029/2011MS000056.
Wang, H., and G. Feingold, 2009a: Modeling mesoscale cellular
structures and drizzle in marine stratocumulus. Part I: Impact
of drizzle on the formation and evolution of open cells.
J. Atmos. Sci., 66, 3237–3256.
——, and ——, 2009b: Modeling mesoscale cellular structures and
drizzle in marine stratocumulus. Part II: The microphysics and
dynamics of the boundary region between open and closed
cells. J. Atmos. Sci., 66, 3257–3275.
——, ——, R. Wood, and J. Kazil, 2010: Modelling microphysical
and meteorological controls on precipitation and cloud cellular structures in southeast Pacific stratocumulus. Atmos.
Chem. Phys., 10, 6347–6362.
Wood, R., 2006: Rate of loss of cloud droplets by coalescence in
warm clouds. J. Geophys. Res., 111, D21205, doi:10.1029/
2006JD007553.
——, M. Köhler, R. Bennartz, and C. O’Dell, 2009: The diurnal
cycle of surface divergence over the global oceans. Quart.
J. Roy. Meteor. Soc., 135, 1484–1493.
——, and Coauthors, 2011: The VAMOS Ocean–Cloud–Atmosphere–
Land Study Regional Experiment (VOCALS-REx): Goals,
VOLUME 69
platforms, and field operations. Atmos. Chem. Phys., 11,
627–654.
Wyant, M. C., C. S. Bretherton, H. A. Rand, and D. E. Stevens,
1997: Numerical simulations and a conceptual model of the
stratocumulus to trade cumulus transition. J. Atmos. Sci., 54,
168–192.
——, and Coauthors, 2010: The PreVOCA experiment: Modeling
the lower troposphere in the southeast Pacific. Atmos. Chem.
Phys., 10, 4757–4774.
Xue, H., G. Feingold, and B. Stevens, 2008: Aerosol effects on
clouds, precipitation, and the organization of shallow cumulus
convection. J. Atmos. Sci., 65, 392–406.
Yuter, S. E., and R. A. Houze Jr., 1995: Three-dimensional kinematic and microphysical evolution of Florida cumulonimbus. Part II: Frequency distribution of vertical velocity,
reflectivity, and differential reflectivity. Mon. Wea. Rev., 123,
1941–1963.
——, Y. L. Serra, and R. A. Houze Jr., 2000: The 1997 Pan
American Climate Studies Tropical Eastern Pacific Process
Study. Part II: Stratocumulus region. Bull. Amer. Meteor. Soc.,
81, 483–490.
Zipser, E. J., 1977: Mesoscale and convective-scale downdrafts as
distinct components of squall-line circulation. Mon. Wea. Rev.,
105, 1568–1589.
Download